Abstract
Objective
Early-life antibiotic exposure, whether through prenatal or childhood antibiotic use, may contribute to increased child body mass. We evaluated associations of prenatal and childhood antibiotic use with body mass index z-score (BMIz) at age three years.
Methods
We utilized electronic health records from 8793 mothers and singleton children delivered at Geisinger Clinic in Pennsylvania, USA, between 2006–2012. Antibiotic orders were ascertained for mothers during pregnancy and for children through their age-three BMI measurement. Linear mixed-effects regression models evaluated associations of prenatal and childhood antibiotic use with child BMIz.
Results
Prenatal antibiotic orders were not associated with child BMIz. Children in the three largest categories of lifetime antibiotic orders had higher BMIz compared to children with no orders; associations persisted when controlling for prenatal antibiotics (beta [95% confidence interval]) (4–5 child orders: 0.090 [0.011, 0.170]; 6–8: 0.113 [0.029, 0.197]; ≥ 9: 0.175 [0.088, 0.263]; trend p-value < 0.001). Two or more first-year orders were also associated with BMIz (1: 0.021 [−0.038, 0.081]; 2: 0.088 [0.017, 0.160]; ≥ 3: 0.104 [0.038, 0.170]; trend p-value < 0.001).
Conclusions
Associations of early-life and lifetime childhood antibiotic use with increased child BMI highlights antibiotic exposure as a modifiable factor for reducing population-level excess weight.
Keywords: Antibiotics, Obesity, Microbiome, Early childhood risk factors
Introduction
Mounting evidence indicates antibiotic use influences childhood weight gain, with implications for the development of obesity. Antibiotics have been recognized as promoting growth in livestock since the 1950s,1 and studies soon followed showing a similar pattern of growth in premature infants treated with antibiotics.2,3 Interest in this line of inquiry has resurfaced alongside increasing efforts to identify determinants of childhood obesity and growing understanding of the gut microbiota’s role in metabolism and weight gain.
Microbial populations in the intestine modulate host metabolism.4,5 Animal and human research indicates gut microbiota affect growth and may play a role in the development of obesity.6,7 Perturbations to gut microbiota that occur early in life appear to disrupt microbial colonization and maturation, with downstream consequences on metabolism.1,8 Antibiotics alter the composition of gut microbiota,9 and while microbial communities generally recover following the discontinuation of antibiotics, some alterations persist.10,11 Even after recovery, research reveals long-term effects on metabolic programming in mice following microbial disruption by antibiotics.12 Thus, antibiotic-induced changes to gut microbiota plausibly influence host metabolism.9
Studies examining childhood antibiotic exposure and body mass report varied findings,13–21 likely due to methodological differences across studies, but the majority demonstrate positive associations. Based on animal studies,12 researchers hypothesize that antibiotic exposure during the first year of life—a critical exposure period in gut microbiota development—may have the greatest impact on later child body mass.1,14 Several of the aforementioned studies report associations between antibiotic exposure in the first year of life and increased body mass later in childhood,13–18,20 although most did not evaluate antibiotic exposure after one year.
Prenatal antibiotic use may influence infant gut microbiota via intergenerational transmission of an antibiotic-altered maternal microbiota, potentially influencing childhood weight gain.22,23 This transfer may occur prior to birth, as emerging evidence reveals the presence of microbes in the intrauterine environment.24 Two studies demonstrate an association between antibiotic use during pregnancy and overweight or obesity in childhood.22,23 Though intriguing, these findings are inconclusive since neither study accounted for childhood antibiotic use. Observed associations between prenatal antibiotic exposure and childhood BMI may be due to an underlying causal relationship between prenatal antibiotic exposure and early childhood antibiotic use.23
Current epidemiological evidence regarding antibiotic exposure and childhood obesity is limited in that no prior study has concurrently examined prenatal and childhood antibiotic use. To address this limitation, we used mother-child linked electronic health record (EHR) data to determine whether prenatal and childhood antibiotic use are independently associated with body mass index (BMI) at age three years. We also compared antibiotic use in the first year of life to lifetime use through age three years. Given prior evidence regarding differential impacts by antibiotic class,15,17 we also evaluated associations by antibiotic class, as well as administration route.
Methods
We conducted a retrospective cohort study using EHR data from the Geisinger Clinic (GC), a large integrated health system in Pennsylvania, USA. GC primary care patients represent the age and sex distribution of the general population in central and northeastern Pennsylvania.25 The Institutional Review Board at the Geisinger Health System approved the study.
Study population
We utilized previously collected EHR data25 from mother-child dyads in which the singleton child was born at GC facilities from 2006–2012; for the current study the child also had to have an age-three BMI measurement. We excluded children with serious birth defects, birth weight < 500 grams, or birth before 22 weeks. From this sample of 9524 dyads, we excluded 731 dyads in which the mother was under age 18 years—given the relationship between younger maternal ages and low birth weight26—or with missing data on pre-gravid BMI or birth weight. Identification of births and linking of mother and child EHRs are previously reported.25
Measures
Child BMIz
The outcome was child BMI z-score (BMIz) at age three years. We selected age three in order to achieve the largest sample of children with a full accounting of early-life antibiotic use while avoiding the inaccuracy of BMI measurements at younger ages.27 Child BMI was calculated from height and weight metrics taken nearest the child’s third birthday (age [years] median: 3.03; interquartile range [IQR]: 2.92, 3.11). The Centers for Disease Control and Prevention 2000 Growth Charts SAS program was used to compute BMIz. BMIz standardizes BMI measurements for sex and age and is reported in standard deviation units.
Antibiotic metrics
Prenatal and childhood antibiotic orders were compiled from medication files. The most common infection diagnoses for antibiotic orders included sinusitis, bronchitis, pharyngitis, acute upper respiratory infections, otitis media, dermatitis, acne, and urinary tract infections, suggesting that antibiotics were prescribed for common, acute infections unlikely to impact birth weight or weight at age three years. Antibiotic orders were used as a surrogate for antibiotic courses because information on dose, duration, and refills is sometimes absent or indiscernible in EHR data. To avoid over-estimating antibiotic courses, if multiple antibiotics were ordered on a given day, only one order was counted. Relevant antibiotics were identified through the Medi-Span Generic Product Identifier Therapeutic Classification System.28
Prenatal antibiotic orders were summed from the conception date through delivery date and analyzed cumulatively and by trimester, with antibiotic orders during labor and delivery (defined as delivery date and day prior to delivery) counted separately from third trimester orders. Childhood antibiotics were analyzed using two methods: lifetime orders (from birth through one day prior to the age three BMI measurement); and orders in the first year of life. To evaluate differences in associations by antibiotic class and administration route, we created counts by the three most common classes of antibiotics in the study population (penicillins, cephalosporins, and macrolides) and an indicator for having received antibiotics non-orally (i.e., intravenous, intramuscular).
Covariates
Child covariates examined include delivery mode (c-section versus vaginal delivery), birth weight tertiles (500–2500, 2500–4000, > 4000 grams), sex, age at the time of age-three BMI measurement (continuous), race/ethnicity (white, black, Hispanic, other, missing), and history of Medical Assistance (yes versus no; used as a proxy for low family socioeconomic status29). Maternal covariates included age at child’s birth (continuous), race/ethnicity, history of Medical Assistance, parity (0, 1, ≥ 2 prior births), smoked during pregnancy (yes versus no), pre-gravid BMI (continuous, centered and centered-squared terms), gestational diabetes (yes versus no), and diabetes mellitus (yes versus no).
Statistical analyses
There were two primary goals of the analysis of antibiotic use in relation to child BMIz at age three years: 1) to evaluate whether there were independent associations of prenatal antibiotic use and lifetime childhood antibiotic use; and 2) to compare early-life use (first year orders) to lifetime use (orders through age three years). For this purpose, linear mixed-effects regression models were used to model child BMIz while controlling for potential confounding variables. A random intercept for mothers was included in models to account for siblings in the sample.
We first evaluated prenatal antibiotic orders (1, 2, ≥ 3 orders versus no orders) and children’s lifetime antibiotic orders (1, 2–3, 4–5, 6–8, ≥ 9 orders versus no orders) separately in unadjusted and adjusted models. Prenatal antibiotics were also evaluated by trimester (≥ 1 order versus no orders) with models for second trimester, third trimester, and labor and delivery antibiotic orders adjusted for earlier prenatal orders. To evaluate independent associations of prenatal and childhood antibiotic exposure with child BMI, we next added prenatal antibiotic orders to the lifetime childhood antibiotic model. Adjusted models controlled for potential confounders of the relation between antibiotic exposure and child BMI, identified a priori, including pre-gravid BMI and smoked during pregnancy, as well as delivery mode and birth weight for childhood antibiotic models (delivery mode and birth weight were not included in adjusted prenatal antibiotics models due to their temporal occurrence after prenatal antibiotic exposure). Models also controlled for child age, race/ethnicity, history of Medical Assistance, and parity. Because associations of primary predictor variables and BMIz did not change with the addition of child sex, maternal age, gestational diabetes, or maternal diabetes, these covariates were excluded from final adjusted models. We conducted extensive model-checking, including the use of added variable plots with a lowess-smooth of the plot.30 Exclusion of influential points (large values for prenatal and childhood antibiotic orders) from models yielded no differences in associations with child BMIz. We also evaluated models with continuous (ln-transformed to reduce the influence of kurtosis) antibiotic order count variables. We evaluated effect modification by pre-gravid BMI (overweight/obese versus normal weight) and delivery mode on antibiotic associations by including cross-product terms of these variables with antibiotic exposure variables. We also evaluated effect modification by birth weight on prenatal antibiotic associations and child sex on childhood lifetime antibiotic associations.
For the other primary analysis goal, we compared associations of childhood antibiotic orders in the first year of life with child BMIz at age three years to those in a model that included orders from the second and third years of life adjusted for first-year orders. Finally, we evaluated antibiotic class by repeating prenatal and lifetime childhood antibiotics models counting the three most common antibiotic classes separately. We evaluated antibiotic administration route by including an indicator for non-oral antibiotic administration.
Results are reported as beta coefficients (representing BMI standard deviation [SD] units) with 95% confidence intervals (CI) and tests for trend p-values. Results were considered significant at P < 0.05 (two-tailed). Stata version 14.0 (StataCorp LP, College Station, TX) was used for data analyses.
Results
Description of mother-child pairs and antibiotic use
The 8793 mothers and three-year-old children in the analysis were primarily white, with substantial proportions receiving Medical Assistance (Table 1). Mean (SD) BMIz among children at age three years was 0.31 (1.15). Among mothers, 60.4% had at least one antibiotic order during pregnancy (Figure 1A) (median: 1; IQR: 0, 2); 33.5% had at least one penicillin order, 22.7% at least one cephalosporin order, and 10.9% at least one macrolide order. Among children, 52.6% had at least one antibiotic order in their first year of life (Figure 1B). By the time of their age three BMI measurement, 82.2% of children had at least one antibiotic order (median: 3; IQR: 1, 6); 76.7% had at least one penicillin order, 38.8% at least one cephalosporin order, and 18.9% at least one macrolide order. There was a significant, although modest, trend of higher lifetime antibiotic use among children with mothers with higher prenatal antibiotic use (Figure 2; Spearman’s rho = 0.123, P < 0.001).
Table 1.
Characteristics of mother-child pairs (n=8793) in study population, by antibiotic exposure status (no antibiotic orders versus at least one order)
| Characteristic | No. (%), unless specified | |
|---|---|---|
| Child characteristics |
No antibiotic orders to age 3 years |
At least one antibiotic order by age 3 years |
| Sex | ||
| Male | 754 (16.6) | 3792 (83.4) |
| Female | 815 (19.2) | 3432 (80.8) |
| Race/ethnicity | ||
| White | 1373 (17.7) | 6399 (82.3) |
| Black | 60 (17.0) | 294 (83.0) |
| Hispanic | 88 (19.3) | 368 (80.7) |
| Other | 40 (24.0) | 127 (76.0) |
| Missing | 8 (18.2) | 36 (81.8) |
| Received Medical Assistance | ||
| Yes | 645 (15.8) | 3442 (84.2) |
| No | 924 (19.6) | 3782 (80.4) |
| Birth weight | ||
| 500–2500 grams | 78 (10.6) | 661 (89.4) |
| 2500–4000 grams | 1366 (18.6) | 5979 (81.4) |
| > 4000 grams | 125 (17.6) | 584 (82.4) |
| Delivery mode | ||
| Caesarian section | 512 (16.3) | 2629 (83.7) |
| Vaginal delivery | 1057 (18.7) | 4595 (81.3) |
| BMI, kg/m2, mean (SD) | 16.35 (1.52) | 16.49 (1.60) |
| BMIz, SD units, mean (SD) | 0.23 (1.14) | 0.33 (1.15) |
| Mother or delivery characteristics |
No antibiotic orders during pregnancy |
At least one antibiotic order during pregnancy |
| Child’s birth weight | ||
| 500–2500 grams | 192 (26.0) | 547 (74.0) |
| 2500–4000 grams | 3005 (40.9) | 4340 (59.1) |
| > 4000 grams | 282 (39.8) | 427 (60.2) |
| Age at child’s birth, mean (SD) | 28.12 (5.58) | 28.04 (5.65) |
| Race/ethnicity | ||
| White | 3145 (39.5) | 4826 (60.5) |
| Black | 96 (35.0) | 178 (65.0) |
| Hispanic | 134 (36.7) | 231 (63.3) |
| Other | 85 (58.2) | 61 (41.8) |
| Missing | 19 (51.4) | 18 (48.6) |
| Received Medical Assistance | ||
| Yes | 1361 (15.5) | 2419 (27.5) |
| No | 2118 (24.1) | 2895 (32.9) |
| Parity | ||
| 0 | 1939 (46.1) | 2272 (53.9) |
| 1 | 927 (35.0) | 1722 (65.0) |
| ≥ 2 | 613 (31.7) | 1320 (68.3) |
| Smoked during pregnancy | ||
| Yes | 670 (34.1) | 1295 (65.9) |
| No | 2809 (41.1) | 4019 (58.9) |
| Pre-pregnancy BMI, kg/m2, mean (SD) | 26.77 (6.85) | 28.23 (7.48) |
| Gestational diabetes | ||
| Yes | 249 (34.0) | 483 (66.0) |
| No | 3230 (40.1) | 4831 (59.9) |
| Diabetes mellitus | ||
| Yes | 43 (22.4) | 149 (77.6) |
| No | 3436 (40.0) | 5165 (60.0) |
Abbreviations: BMI = body mass index; BMIz = body mass index z-score; SD = standard deviation
Figure 1.
Antibiotic orders in women during pregnancy (A) and in childrena (B)
a Lifetime through age 3 period includes antibiotic orders from birth through the day before the child’s age-three BMI measurement
Figure 2.
Proportion of children in each category of lifetime antibiotic ordersa by category of prenatal antibiotic exposure among their mothers
a Includes antibiotic orders from birth through the day before the child’s age-three BMI measurement
Prenatal and lifetime childhood antibiotic use and child BMIz
In unadjusted analysis, only children whose mothers had three or more prenatal antibiotic orders had significantly higher BMIz compared to children of mothers with no orders (beta [CI]) (1 order: 0.027 [−0.028, 0.083]; 2 orders: 0.034 [−0.037, 0.105]; ≥ 3 orders: 0.117 [0.039, 0.194]; trend P = 0.006). This association was no longer present after adjustment for covariates (Table 2; trend P = 0.30). When prenatal antibiotic orders were modeled as a continuous variable, there was still no association with BMIz (P = 0.39). Analysis of prenatal antibiotics by trimester revealed no associations with BMIz (results not shown).
Table 2.
Adjusteda association of prenatal antibiotic use with child BMI z-score at age 3 years
| Variable | Beta, SD units (CI) | P-value |
|---|---|---|
| Prenatal antibiotic orders, ref = 0 |
||
| 1 | 0.007 (−0.048, 0.063) | 0.80 |
| 2 | 0.000 (−0.070, 0.071) | 0.99 |
| ≥ 3 | 0.050 (−0.027, 0.128) | 0.20 |
Abbreviations: SD = standard deviation; CI = confidence interval; ref = reference group
Multivariable linear regression model controlled for centered child exact age, mother race/ethnicity, mother Medical Assistance, smoked during pregnancy, parity, and pre-gravid BMI (centered and centered-squared terms), as described in the Methods; each was significantly associated with child BMIz.
In unadjusted and adjusted analyses, children in the three highest categories of lifetime antibiotic orders had significantly higher BMIz at age three years compared to children with no orders, with a trend of increasing BMIz across categories (unadjusted results: not shown; adjusted results: Table 3, Model 2a; trend P < 0.001). There was a significant trend of increasing BMI with increasing childhood antibiotic use in a model that included antibiotic orders as a continuous variable (P < 0.001). Associations persisted after adjustment for prenatal antibiotic orders (Table 3, Model 2b; trend P < 0.001).
Table 3.
Adjusteda association of lifetime child antibiotic use through age 3 and child BMI z-score at age 3 years, without and with prenatal antibiotic use in the model
| Variable | Model 2a: Lifetime child antibiotics only |
Model 2b: Lifetime child antibiotics, adjusted for prenatal antibiotics |
||
|---|---|---|---|---|
| Beta, SD units (CI) | P-value | Beta, SD units (CI) | P-value | |
| Lifetime child antibiotic orders through age 3 BMI measurement, ref = 0 |
||||
| 1 | 0.036 (−0.043, 0.115) | 0.37 | 0.035 (−0.044, 0.114) | 0.39 |
| 2–3 | 0.023 (−0.049, 0.095) | 0.53 | 0.021 (−0.051, 0.093) | 0.58 |
| 4–5 | 0.093 (0.014, 0.172) | 0.02 | 0.090 (0.011, 0.170) | 0.03 |
| 6–8 | 0.117 (0.033, 0.200) | 0.01 | 0.113 (0.029, 0.197) | 0.008 |
| ≥ 9 | 0.183 (0.096, 0.270) | < 0.001 | 0.175 (0.088, 0.263) | < 0.001 |
Abbreviations: SD = standard deviation; CI = confidence interval; ref = reference group
Multivariable linear regression models controlled for caesarian section, birth weight (tertiles), centered child exact age, child race/ethnicity, child Medical Assistance, smoked during pregnancy, parity, and pre-gravid BMI (centered and centered-squared terms), as described in the Methods; each was significantly associated with child BMIz.
Early life antibiotic use and child BMIz
Children with two or more antibiotic orders in the first year of life had significantly higher BMIz at age three years compared to children with no orders, with a trend of increasing BMIz across categories (beta [CI]) (1 order: 0.021 [−0.038, 0.081]; 2 orders: 0.088 [0.017, 0.160]; ≥ 3 orders: 0.104 [0.038, 0.170]; trend P < 0.001). There was a significant trend of increasing BMIz with increasing antibiotic use in a model that included first-year orders as a continuous variable (P = 0.002). In a model with second and third year antibiotic orders adjusted for first-year orders, children with six or more antibiotic orders had significantly higher BMIz at age three years compared with children with no orders (1 order: 0.015 [−0.054, 0.084; 2–3 orders: 0.053 [−0.012, 0.118]; 4–5 orders: 0.054 [−0.024, 0.133]; ≥ 6 orders: 0.123 [0.042, 0.204]; trend P = 0.003).
Sensitivity and secondary analyses
Given our exclusion criteria, the study population comprised largely healthy children; however, to explore whether low birth weight children potentially altered associations we excluded 739 low birth weight children (500–2500 grams) from the adjusted antibiotics models. Exclusion of low birth weight children from prenatal and child lifetime antibiotics models did not change inferences.
Evaluation of effect modification by pre-gravid BMI resulted in a significant interaction term for the largest category of prenatal antibiotic use, suggesting that among mothers categorized as overweight/obese, three or more antibiotic orders during pregnancy may be associated with increased child BMIz at age three years (≥ 3 orders: 0.136 [−0.144, 0.417]). Neither birth weight nor delivery mode modified the relation of prenatal antibiotic use with child BMIz. There was no evidence that delivery mode modified associations of lifetime childhood antibiotic orders and child BMIz when antibiotic use was included as a categorical variable; however, when included as a continuous variable, the association with BMIz was stronger in children delivered vaginally (P = 0.005). Neither pre-gravid BMI nor sex modified the relation of lifetime childhood antibiotic use with child BMIz.
In analyses by antibiotic class, there were no significant associations between prenatal orders of any class and child BMIz (results not shown). In models of lifetime childhood antibiotic use, the highest category of antibiotic orders for each class (which differed based on the prevalence of use of each antibiotic) was significantly associated with child BMIz, adjusting for orders from other classes (penicillins, ≥ 4 orders: 0.079 [0.007, 0.150], trend P across all categories = 0.076; macrolides, ≥ 2 orders: 0.102 [0.011, 0.192], trend P across all categories = 0.041; cephalosporins, ≥ 3 orders: 0.112 [0.029, 0.196], trend P across all categories = 0.024). To compare associations across classes, we re-categorized antibiotic orders by class (0, 1, ≥ 2 orders) and repeated analyses. The most robust association was for macrolides, which was 35% larger than cephalosporins and 77% larger than penicillins. This comparison likely underestimated the relative strength of the macrolide association, as the categories for two or more orders of cephalosporins and penicillins contained many more instances of three or more antibiotic orders as compared to macrolides. Antibiotic administration route among mothers or children was not associated with child BMIz (results not shown).
Discussion
In this retrospective cohort study of 8793 mother-child pairs, the first study to evaluate independent contributions of prenatal and childhood antibiotic use, we found that both lifetime and first-year childhood antibiotic use was associated with BMI at age three years, but—in contrast to prior studies22,23—prenatal antibiotic use was not. The association between lifetime childhood antibiotic use and child BMI did not attenuate after controlling for prenatal antibiotic orders, indicating that the role of childhood antibiotic use on child BMI was independent of prenatal antibiotic exposure. We also observed that macrolides were more strongly associated with child BMI than were cephalosporins or penicillins.
We did not observe an association between prenatal antibiotic orders and child BMI at age three years, either cumulatively or by trimester, although three or more prenatal antibiotic orders among mothers categorized as overweight/obese was associated with higher child BMI in a post hoc analysis. Our findings differ from two prior studies that observed associations between prenatal antibiotic exposure in the second and third trimesters of pregnancy and increased child BMI.22,23 The difference in our findings are likely due to methodological differences; for example, these prior studies assessed child BMI at older ages (ages seven to 16), one study lacked information on pre-gravid BMI, a likely confounder of the relation between prenatal antibiotics and child BMI,22 and one study with a relatively small sample size relied on questionnaires to ascertain prenatal antibiotic use, and is thus subject to recall bias.23
Higher lifetime childhood antibiotic use was associated with higher child BMI at age three. The magnitude of associations of the three highest categories of lifetime childhood antibiotics with child BMI were comparable to associations we observed with c-section delivery, receiving Medical Assistance (a proxy for low socioeconomic status), and having smoked during pregnancy (results not shown), respectively, known and important risk factors for obesity.31,32 Similar to past studies,15,17,20 we also observed a dose-effect relation between lifetime childhood antibiotic use and child BMI, with associations starting at four antibiotic orders. Considering the majority of studied children received at least four orders by age three years, the associations we observed with child BMI are concerning for population-level excess weight (that is, the number of children in the overweight and obese tail of the BMI distribution after the shift in the population BMI distribution to the higher values reflected in our data).
Consistent with several prior studies,14–18 we found that, compared to children with no antibiotic orders, children with two or more orders in the first year of life had higher BMI at age three years, again with a dose-effect relation. In contrast, a study by Gerber et al. reported no association between antibiotic exposure in the first six months of life and weight gain in children through age seven years, though a significant, dose-effect relation with antibiotic exposure in the first 24 months of life.19 While difficult to make direct comparisons due to differences in study methods, disparities in early-life antibiotic exposure between the two study populations may explain the discrepancy in our findings. Most (79%) of the exposed six-month-old children in Gerber et al.’s study received just one antibiotic course, compared to 42% of the exposed in our sample. We did not find associations between single first-year antibiotic orders and child BMIz; rather, associations began at two orders and strengthened with greater antibiotic exposure.
Compared to antibiotic orders that occurred after year one, first-year orders were associated with child BMI at a lower number of orders, providing further support that the first year of life may represent a critical exposure period for antibiotics to alter children’s developing gut microbiota.1,14 However, even after adjustment for first-year orders, associations with antibiotic orders in years two and three remained robust for the largest category of antibiotic use, indicating that cumulative childhood antibiotic exposure plays an important role in childhood weight gain. These results build upon a study by our group that utilized child-only EHR data and showed cumulative and persistent effects of antibiotics on BMI throughout childhood.21 Evidence suggests that repeated antibiotic use may not allow gut microbiota to recover from antibiotic-induced perturbations, resulting in a persistent antibiotic-altered microbiota composition.11
Consistent with past studies,11,15,17 compared with cephalosporins and penicillins, macrolides had the most robust association with child BMI. Macrolide use in childhood alters intestinal microbiota in ways that predispose children to weight gain, including reducing bacterial diversity, increasing the abundance of endotoxin-producing organisms, and depleting beneficial bacteria and organisms that play a key role in metabolism, changes not observed with penicillin use.11
This is the first study of antibiotic use and child body mass to utilize mother-child linked EHR data, which allowed for concurrent evaluation of prenatal and childhood antibiotic exposure. Additional strengths include the use of EHR data, providing more accurate characterizations of antibiotic use than self-report or parental recall, and a full accounting of antibiotic orders in children from birth through the time of BMI measurement. However, we lacked information on whether patients took antibiotics as prescribed, nor could we fully ascertain antibiotic prescriptions that occurred outside the GC. In either case, associations may be underestimated due to regression dilution. Antibiotic exposure metrics were also limited by not factoring in dose or duration; however, we incorporated information on antibiotic class and administration route, overcoming limitations cited by prior studies.16,23 Finally, infection status could confound associations between antibiotic use and BMI; however, diagnoses for antibiotic orders largely comprised infections unlikely to impact weight.
Conclusion
In this first study to concurrently evaluate associations of prenatal and childhood antibiotic use with child BMI, we found that children’s antibiotic use—and not prenatal antibiotic exposure—is related to children’s BMI at age three years. Associations between antibiotic use in the first year of life and child BMI support growing evidence that early-life may represent a critical exposure period; however, the relation between antibiotic use and child BMI is not limited to this first year. Strong associations between lifetime antibiotic orders and child BMI highlight the potential risk associated with cumulative exposure to antibiotics in childhood. By avoiding use of broad-spectrum antibiotics when feasible,15 limiting repeated antibiotic exposure,17 eliminating antibiotic use in cases that lack evidence of efficacy, and recognizing the role of cumulative antibiotic exposure in weight gain throughout childhood,21 this common, population-wide exposure presents a modifiable factor for reducing obesity risk.
What’s known on this subject
Early childhood use of antibiotics is associated with increased child body mass
Initial studies suggest a similar relationship with prenatal antibiotic use, but do not account for childhood antibiotic exposure
What this study adds
This is the first study to concurrently evaluate both prenatal and childhood antibiotic use in relation to child body mass index
Childhood antibiotic exposure, but not prenatal antibiotic exposure, was associated with increased body mass at age three years
Both antibiotic exposure in the first year of life and cumulative exposure over the first three years of life were associated with increased body mass at age three years
Acknowledgments
Funding source: Research reported in this publication was supported by the Global Obesity Prevention Center (GOPC) at Johns Hopkins, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) and the Office of the Director, National Institutes of Health (OD) under award number U54HD070725. The funders had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflicts of interest: The authors received grant funding from the National Institutes of Health, which was used to support this study.
References
- 1.Cox LM, Blaser MJ. Antibiotics in early life and obesity. Nat Rev Endocrinol. 2015;11(3):182–190. doi: 10.1038/nrendo.2014.210. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Perrini F. Aureomycin as growth factor in prematures. Boll Soc Ital Biol Sper. 1951;27(7–8):1151–1152. [PubMed] [Google Scholar]
- 3.Robinson P. Controlled trial of aureomycin in premature twins and triplets. Lancet. 1952;259:52. [Google Scholar]
- 4.Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, Gordon JI. An obesity-associated gut microbiome with increased capacity for energy harvest. Nature. 2006;444(7122):1027–1031. doi: 10.1038/nature05414. [DOI] [PubMed] [Google Scholar]
- 5.Jumpertz R, Le DS, Turnbaugh PJ, et al. Energy-balance studies reveal associations between gut microbes, caloric load, and nutrient absorption in humans. Am J Clin Nutr. 2011;94(1):58–65. doi: 10.3945/ajcn.110.010132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Angelakis E, Merhej V, Raoult D. Related actions of probiotics and antibiotics on gut microbiota and weight modification. Lancet Infect Dis. 2013;13(10):889–899. doi: 10.1016/S1473-3099(13)70179-8. [DOI] [PubMed] [Google Scholar]
- 7.Sanmiguel C, Gupta A, Mayer EA. Gut Microbiome and Obesity: A Plausible Explanation for Obesity. Curr Obes Rep. 2015;4(2):250–261. doi: 10.1007/s13679-015-0152-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mueller NT, Bakacs E, Combellick J, Grigoryan Z, Dominguez-Bello MG. The infant microbiome development: mom matters. Trends Mol Med. 2015;21(2):109–117. doi: 10.1016/j.molmed.2014.12.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Mikkelsen KH, Allin KH, Knop FK. Effect of antibiotics on gut microbiota, glucose metabolism and bodyweight regulation - a review of the literature. Diabetes Obes Metab. 2016;18(5):444–453. doi: 10.1111/dom.12637. [DOI] [PubMed] [Google Scholar]
- 10.Jernberg C, Lofmark S, Edlund C, Jansson JK. Long-term impacts of antibiotic exposure on the human intestinal microbiota. Microbiology. 2010;156(Pt 11):3216–3223. doi: 10.1099/mic.0.040618-0. [DOI] [PubMed] [Google Scholar]
- 11.Korpela K, Salonen A, Virta LJ, et al. Intestinal microbiome is related to lifetime antibiotic use in Finnish pre-school children. Nat Commun. 2016;7:10410. doi: 10.1038/ncomms10410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Cox LM, Yamanishi S, Sohn J, et al. Altering the intestinal microbiota during a critical developmental window has lasting metabolic consequences. Cell. 2014;158(4):705–721. doi: 10.1016/j.cell.2014.05.052. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ajslev TA, Andersen CS, Gamborg M, Sorensen TI, Jess T. Childhood overweight after establishment of the gut microbiota: the role of delivery mode, pre-pregnancy weight and early administration of antibiotics. Int J Obes (Lond) 2011;35(4):522–529. doi: 10.1038/ijo.2011.27. [DOI] [PubMed] [Google Scholar]
- 14.Azad MB, Bridgman SL, Becker AB, Kozyrskyj AL. Infant antibiotic exposure and the development of childhood overweight and central adiposity. Int J Obes (Lond) 2014;38(10):1290–1298. doi: 10.1038/ijo.2014.119. [DOI] [PubMed] [Google Scholar]
- 15.Bailey LC, Forrest CB, Zhang P, Richards TM, Livshits A, DeRusso PA. Association of antibiotics in infancy with early childhood obesity. JAMA Pediatr. 2014;168(11):1063–1069. doi: 10.1001/jamapediatrics.2014.1539. [DOI] [PubMed] [Google Scholar]
- 16.Murphy R, Stewart AW, Braithwaite I, et al. Antibiotic treatment during infancy and increased body mass index in boys: an international cross-sectional study. Int J Obes (Lond) 2014;38(8):1115–1119. doi: 10.1038/ijo.2013.218. [DOI] [PubMed] [Google Scholar]
- 17.Saari A, Virta LJ, Sankilampi U, Dunkel L, Saxen H. Antibiotic exposure in infancy and risk of being overweight in the first 24 months of life. Pediatrics. 2015;135(4):617–626. doi: 10.1542/peds.2014-3407. [DOI] [PubMed] [Google Scholar]
- 18.Trasande L, Blustein J, Liu M, Corwin E, Cox LM, Blaser MJ. Infant antibiotic exposures and early-life body mass. Int J Obes (Lond) 2013;37(1):16–23. doi: 10.1038/ijo.2012.132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Gerber JS, Bryan M, Ross RK, et al. Antibiotic exposure during the first 6 months of life and weight gain during childhood. JAMA. 2016;315(12):1258–1265. doi: 10.1001/jama.2016.2395. [DOI] [PubMed] [Google Scholar]
- 20.Scott FI, Horton DB, Mamtani R, et al. Administration of antibiotics to children before age 2 years increases risk for childhood obesity. Gastroenterology. 2016;151(1):120–129. doi: 10.1053/j.gastro.2016.03.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Schwartz BS, Pollak J, Bailey-Davis L, et al. Antibiotic use and childhood body mass index trajectory. Int J Obes (Lond) 2015;40(4):615–621. doi: 10.1038/ijo.2015.218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Mor A, Antonsen S, Kahlert J, et al. Prenatal exposure to systemic antibacterials and overweight and obesity in Danish schoolchildren: a prevalence study. Int J Obes (Lond) 2015;39(10):1450–1455. doi: 10.1038/ijo.2015.129. [DOI] [PubMed] [Google Scholar]
- 23.Mueller NT, Whyatt R, Hoepner L, et al. Prenatal exposure to antibiotics, cesarean section and risk of childhood obesity. Int J Obes (Lond) 2015;39(4):665–670. doi: 10.1038/ijo.2014.180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Matamoros S, Gras-Leguen C, Le Vacon F, Potel G, de La Cochetiere MF. Development of intestinal microbiota in infants and its impact on health. Trends Microbiol. 2013;21(4):167–173. doi: 10.1016/j.tim.2012.12.001. [DOI] [PubMed] [Google Scholar]
- 25.Casey JA, Savitz DA, Rasmussen SG, et al. Unconventional Natural Gas Development and Birth Outcomes in Pennsylvania, USA. Epidemiology. 2016;27(2):163–172. doi: 10.1097/EDE.0000000000000387. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Martin JA, Hamilton BE, Osterman MJK, Curtin SC, Matthews TJ. Births: Final data for 2013. Division of Vital Statistics, Centers for Disease Control and Prevention; 2015. [PubMed] [Google Scholar]
- 27.Foote JM, Brady LH, Burke AL, et al. Development of an evidence-based clinical practice guideline on linear growth measurement of children. J Pediatr Nurs. 2011;26(4):312–324. doi: 10.1016/j.pedn.2010.09.002. [DOI] [PubMed] [Google Scholar]
- 28.Medi-Span. Master Drug Data Base Documentation Manual. Indianapolis, IN, USA: Medi-Span; 2007. [Google Scholar]
- 29.Schwartz BS, Bailey-Davis L, Bandeen-Roche K, et al. Attention deficit disorder, stimulant use, and childhood body mass index trajectory. Pediatrics. 2014;133(4):668–676. doi: 10.1542/peds.2013-3427. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bandeen-Roche K, Hall CB, Stewart WF, Zeger SL. Modelling disease progression in terms of exposure history. Statistics in medicine. 1999;18(21):2899–2916. doi: 10.1002/(sici)1097-0258(19991115)18:21<2899::aid-sim203>3.0.co;2-f. [DOI] [PubMed] [Google Scholar]
- 31.Kuhle S, Tong OS, Woolcott CG. Association between caesarean section and childhood obesity: a systematic review and meta-analysis. Obes Rev. 2015;16(4):295–303. doi: 10.1111/obr.12267. [DOI] [PubMed] [Google Scholar]
- 32.Woo Baidal J, Locks L, Cheng E, Blake-Lamb T, Perkins M, Taveras E. Risk factors for childhood obesity in the first 1,000 days: A systematic review. American Journal of Preventive Medicine. 2016;50(6):761–779. doi: 10.1016/j.amepre.2015.11.012. [DOI] [PubMed] [Google Scholar]